Our paper on improving clean-label backdoor attacks in constrained threat models is accepted at ICLR 2025.